The fact you mention $__interval suggests that you're asking a question about Grafana, rather than Prometheus.
Maybe $__rate_interval would help you? https://grafana.com/blog/2020/09/28/new-in-grafana-7.2-__rate_interval-for-prometheus-rate-queries-that-just-work/ As long as you set the sampling rate correctly in the data source definition, then this will put at least 2 data points in the window, and up to 4 when you zoom in. Hence if you have one missing point, you'll still get at least one data point for avg_over_time(). On Thursday, 2 June 2022 at 18:53:45 UTC+1 [email protected] wrote: > H! I have statsd_exporter witch provide me honest 95 percentile for every > api endpoint in window of 15 sec. > > I am also trying to visualize it in this type of chart: > > [image: Screen Shot 2022-05-28 at 06.10.32.png] > Every block have represent time window with will expense if I zoom out > time filter. > I decide to show avg between calculated 15s percentiles for longer > $__interval. > > I come out with PromQL below, but found out that if for api_object in my > $__interval window one point is missing then avg_over_time return null for > whole interval. > sum_over_time and quantile_over_time works the same, but max_over_time and > min_over_time returns good value in this case. > > avg by (api_object) (avg_over_time(vendor_api_request{quantile="0.95", > env="$env", http_method="GET"}[$__interval])) > > Could I got avg over time for metric with absent of some values (nulls)? > [image: Screen Shot 2022-05-28 at 06.27.55.png] > > I am aware of histograms and really interested of avg between percentiles > in this case. > > > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/5efa8163-e3f4-4eb5-9a90-326a330ca193n%40googlegroups.com.
